Recently, microplastics (MP) have emerged as global contaminants that seriously affect human and ecological health. However, rapid identification of MP is still a challenge, whether from oceans, wastewater, sediment or soil. A system based on laser-Raman spectroscopy analysis for qualitative testing of MP was established. The monitoring system can realize in-situ real-time detection and nondestructive testing, which provide a large amount of Raman spectroscopy of MP for Marine environmental analysis. A database suitable for microplastics analysis was presented based on the characteristic of Raman spectroscopy. Extra Trees algorithm was presented for the automatic identification of MP in this paper. The algorithm network is trained to detect random MP based on the established database, which including pure MP and mixed MP. The experiment result shows that several MP samples, including pure polystyrene (PS), Polymethyl methacrylate (PMMA), polyethylene terephthalate (PET), polyethylene (PE), Polyamide (PA), polyvinyl chloride (PVC) and polypropylene (PP) could be individually and automatically identified. The experiment result demonstrated that over 98.82% mixed particles could be correctly identified. The results were consistent with Extra Trees model built for identifying six types of MP, indicating Extra Trees model was highly robust for more than six of MP detection. The spectroscopy analysis method in this paper provides data support for systematically understanding the microplastic contamination.